A few months ago, we published a roundup of the most recent and rigorous research into the effects of police de-escalation programs on civilian and officer safety. Our roundup was prompted by the Law Enforcement De-escalation Training Act of 2022 (Public Law No. 117-325), through which Congress authorized the appropriation of $124 million over the next four years for the development, implementation, and study of training curricula meant to reduce the use of lethal force in interactions between law enforcement personnel and the public. As we previously noted, LEDTA is important for two reasons: first, the scale of the proposed investment is unusually large and represents a new funding stream for police activities, one that can and will be evaluated on its own over time. Second, LEDTA is meaningful bipartisan reform legislation and a reminder that federal-led justice innovation is still possible.

LEDTA is not the comprehensive, top-down police reform bill many progressives sought. Still, it accomplishes much. In our previous roundup, we focused on where the bill most directly addressed police-involved killings: de-escalation of potentially violent situations between police officers and unarmed suspects who pose little risk to public safety. LEDTA’s structure puts these confrontations front and center, while also encompassing a set of prevalent, fraught, and often deadly interactions: mental and behavioral health crises. 

In this roundup, we will address the broader and more methodologically heterodox literature that deals with police responses to such events. 

Situations that involve people in behavioral crises (PBCs) comprise a much wider category of events than those that require mere de-escalation. Implicit in the distinction between de-escalation and mental health responses is both a question of culpability and an increased chance of danger. In the former, emotion and miscommunication on the part of both police and civilians cause injuries and death. Mental health crises, meanwhile, result from underlying issues that may require more specialized responses than police are prepared to offer, even with the best training. What’s more, such crises may require not just “heat-of-the-moment” interventions to defuse tempers but also long-term prevention and care to make future incidents less likely. 

LEDTA prioritizes the moment of confrontation, ostensibly because that is where improved police training might have the biggest impact. Though such confrontations exist on a continuum that includes many chances for intervention, the act focuses on the flashpoints, defining mental health crises as events in which the “behavior of the person puts the person at risk of hurting himself, herself, or others; or impairs or prevents the person from being able to care for himself or herself or function effectively in the community.” The definition includes instances of acute alcohol or drug misuse, suicide attempts, and other situations in which individuals display “symptoms of mental illness or…psychological impairments.” Clearly, no one curriculum can fit the breadth of these contingencies. 

Just as clear from the research is that mental health crises currently consume a significant portion of police resources. Depending on region and locality, between 7 and 31 percent of police calls involve someone with symptoms of a mental or behavioral health disorder. Despite this, people with generalized mental health disorders are not significantly more likely to commit violent crimes than the rest of the population, and even people with serious mental illnesses are relatively unlikely to commit a random violent act. Among patients discharged from inpatient facilities after mental health crises, only about 2 percent committed a subsequent violent act involving a gun in the year following discharge, and only 6 percent committed any type of violent act involving a bystander. 

Still, some specific mental and behavioral disorders are more likely to lead to violence. Serious and persistent mental illness such as schizophrenia accounts for about 4 percent of all reported incidents of violence despite a population prevalence of only about 0.32 percent. (Schizophrenia is also overrepresented in U.S. state prisons; an estimated 2 to 6.5 percent of state prisoners experience schizophrenia.) What’s more, substance abuse exacerbates the risks inherent in underlying mental illness. Lifetime risk to commit violence was approximately 33 percent among those with schizophrenia and other similarly serious mental illnesses and 55 percent among people with both mental illness and substance abuse issues, compared to 15 percent among people without either diagnosis. As we have noted before, alcohol misuse disorders are significantly associated with violent crime, suicide, and a host of other dangerous outcomes.

Several factors help explain why PBCs are more likely to find themselves in violent confrontations with police. They are far more likely than the rest of the population to be criminally victimized, which as Harold Pollack and Keith Humphreys point out, means they have more contact with police at times when they are distressed and vulnerable. The locations at which PBCs interact with police often involve fraught interpersonal dynamics. Pollack and Humphreys note that the most common setting for a behavioral crisis is at home, and these events often involve romantic partners, caregivers, or other family members. Family conflict calls are notably dangerous, as they are often made long after situations have gotten out of control and involve vulnerable parties, such as children and the elderly. These factors mean that police are more likely to resort to force in response to potential violence than they might otherwise be. 

The frequency and nature of interactions between PBCs and police contribute to negative outcomes. People with serious mental illness are vastly overrepresented in prison and jail populations, and corrections facilities are often ill-equipped to fill the role of mental and behavioral health service providers, leading to institutional violence and frequent recidivism among people with mental and behavioral health challenges. Interactions between PBCs and police are characterized by violence, both in the U.S. and abroad. A 2016 study of police interactions in New Zealand, a country known for low levels of violent crime, found that police used force in 78% of interactions with PBCs. Mental illness was a factor in 25 percent of fatal police shootings in the U.S. reviewed from 2015 to 2018. Some estimates suggest that people with a serious mental illness are up to 16 times more likely to be killed in a police interaction than the rest of the population.

Since police intervention is the most frequent response to situations in which PBCs are at risk of violence, LEDTA’s emphasis on police training to prepare for these situations makes sense. The most prominent and best-studied mental health response curriculum for police officers is the Crisis Intervention Team (CIT) model, pioneered by the Memphis Police Department. In the more than 30 years since its development, researchers have consistently found that, unlike de-escalation, CIT does not reduce the use of force by police on civilians. Still, the other outcomes of these interactions do seem better among CIT-trained officers. 

In 2010, for example, Joshua Acker found no difference in the use of force among CIT officers compared to their untrained colleagues, but the likelihood of civilian injury was significantly smaller. Similarly, in 2014 Michael Compton, and colleagues found that CIT officers used force just as often as their non-CIT colleagues. However, they were more likely to employ low-level force coupled with verbal engagement and negotiation. 

Moreover, many of the other benefits attributable to CIT seem concentrated among civilians for whom officers are most likely to see the use of force as necessary. Officers with CIT training were more likely to transport PBCs to mental health services and less likely to arrest PBCs, even when force was used. Similarly, in 2006, Jennifer Teller and colleagues found that CIT officers were more likely to resolve conflicts at the scene or transport PBCs to emergency psychiatric services. In 2019 Erin Comartin and colleagues found that not only were CIT officers more likely to transport PBCs to crisis centers, but that they were more willing to go out of their way to do so. 

The willingness of CIT officers to defer PBCs to psychiatric services speaks to the growing realization that police are not best equipped to deal with mental and behavioral health crises. Another form of response couples the coercive and protective functions of law enforcement with trained specialists, a model known as co-response. Co-response models typically include one or more police officers, a mental health clinician, and sometimes a paramedic or nurse working in coordination at the scene. Co-response has not been extensively evaluated in the U.S., but recent international studies suggest that the practice has benefits in specific applications. 

Susanna Every-Palmer and colleagues evaluated a co-response program in New Zealand in 2022. Though they found no reduction in the use of force, they did find a significant reduction in emergency department (ED) referrals, suggesting that co-response lowers the chance of injury or prolonged psychiatric episodes. In 2022, Etienne Blais and colleagues evaluated Mobile Crisis Intervention Teams (MCIT), which pair officers with psychiatric nurses in the field, and found that in Canada, they are associated with more referrals to community services and fewer involuntary hospitalizations and uses of force than were police acting on their own. 

Finally, Blais and David Brisebois evaluated a version of co-response that connects police responding to attempted suicides with offsite clinicians who helped them evaluate the situation and develop a plan of action. They found that these co-responses were associated with more referrals to community services and fewer involuntary hospitalizations. They also found a slightly lower rate of use of force among co-responding officers. 

The evaluations reviewed here imply that the use of force may not be the only or even primary outcome of interest we should be worried about regarding responses to PBCs. Often, when the elements of de-escalation (space, time, and cover) fail, force must be used to prevent harm to self and others. Instead, what type of force police employ and what happens after force is applied may matter more. The best available evidence suggests that responses to PBCs informed by specialized curricula may not reduce the use of force but improve the incidence of injuries, the likelihood of arrest, and the chance that PBCs will get the aftercare they need to avoid police confrontations in the future. Violence is relatively concentrated among people with mental and behavioral health challenges, and the hope is that as more PBCs get help, police can concentrate their resources on serving only those who are most at-risk and pose the most danger to others.

Police confrontations are part of a behavioral continuum along which there are many opportunities for intervention. Responses to PBCs are often perilous, and police managers are understandably unwilling to leave their officers’ responses up to chance. The research reviewed here focuses on heat-of-the-moment tactics meant for situations in which police have little room for failure. For this reason, the evidence is relatively slim. None of the available research employs the randomization of treatment necessary to make causal claims about the effects of the training programs, despite the long history of mental and behavioral health training among police. As a result, those tasked with developing training curricula under LEDTA must make do with what’s available to them. The law’s broad definition of “mental and behavioral crisis” should allow police agencies and the policies that govern them to spread their resources to prevention and aftercare, taking some pressure off responding officers. Moreover, its mandate for evaluation means that we will soon know more about how these resources can be most effectively allocated. When we do, we can paint a much clearer picture of the role training plays in saving the lives of people caught in crisis.